Search Menu
Home Latest News Menu

Musical capability may be an instinctive trait, new study suggests

Researchers from KAIST break down musical instincts with AI

  • 7 May 2024

Music is acknowledged as a pervasive element across global cultures, often described as the universal language. This raises the question: Is there a "musical instinct" inherent across humanity despite the varied environmental backgrounds?

A research team from KAIST (The Korea Advanced Institute of Science and Technology) led by Professor Hawoong Jung of the Department of Physics announced a groundbreaking discovery. They've pinpointed the principle through which musical instincts emerge in the human brain. Remarkably, this is achieved without formal musical training, using an artificial neural network model.

Historically, researchers have delved into the commonalities and distinctions in music across different cultures, aiming to decode its universal appeal. A 2019 study confirmed that all distinct cultural groups create music, often utilising similar beats and melodies. Neuroscientists have also identified the auditory cortex as the brain region responsible for processing musical sounds.

Professor Jung's team employed an artificial neural network, demonstrating that cognitive functions for music develop naturally through the processing of auditory information from the environment, without the need to learn music explicitly. The research utilised Google's AudioSet, a vast sound data collection, training the network to differentiate various sounds. Intriguingly, the team discovered that certain neurons within the network responded selectively to music. These neurons showed minimal reaction to other sounds like those from animals or machines but were highly responsive to different types of music, both instrumental and vocal.

These neurons mirrored the response patterns seen in the auditory cortex of an actual brain. For example, they showed decreased response to music that was fragmented into short segments and rearranged, suggesting these neurons encode music's temporal structure. This characteristic spanned across 25 genres, from classical to electronic.

Moreover, inhibiting these music-selective neurons significantly reduced the network's accuracy in processing other natural sounds, suggesting that musical capability might be an instinctive trait evolved to enhance the processing of environmental sounds.

Professor Jung commented, “The results of our study imply that evolutionary pressure has contributed to forming the universal basis for processing musical information in various cultures. We look forward for this artificially built model with human-like musicality to become an original model for various applications including AI music generation, musical therapy, and for research in musical cognition.”

Find out more about the findings here.

Load the next article